To be specific, I am analyzing a phonological process like this

/maNioN/ --> [mãn.i.õn] --> [mã.ni.õn] --> [mã.nyõ?]

These intermediary forms make sense according to the phonology of the language. Does the transcription make sense, i.e. having intermediary forms between brackets ? Is it even correct to have intermediary forms? I guess not in OT, but maybe it's tolerated in a rule-based approach, or is it?


2 Answers 2


I think a question about "standard" is asking for trouble, since there's no definition of "what's standard" and no behavioral test for discerning if something is "standard". That use of square brackets is, nevertheless, rather at odds with historical professional practice in generative phonology. It may be observed contemporarily in homeworks in phonology classes, where students don't know what is standard and instructors may not feel like making a point of objecting to use of square brackets in pre-surface forms. In journal publications, it is difficult these days to find derivations where you could test what is standard, for two reasons. Most obviously, publications in phonology over the past 20 years which include underlying-to-surface information are usually framed in OT which uses the finger to identify surface forms. The second reason is that square brackets were only used to (completely) enclose segment sequences, but derivations under autosegmental phonology became representationally richer, including all sorts of add-on tiers which were not enclosed in square brackets. So original observed standards regarding use of square brackets fell into desuetude.

If you focus on papers written up to the 80's, then you will find that square brackets are reserved for the actual output. See Hyman's textbook p. 82, 85, 87 etc. Even then, there was a lot of variation, though not to the point of putting "surface form" brackets around every intermediate representation. Note that square brackets were also used to indicate cyclic bracketing and morphological structure. It was also common to not use square brackets at all in a derivation, because the surface form was the last form in a derivation (see Hyman's textbook p. 128, 130 etc).

There is a typographic desideratum in publications to the effect that linguistic data has to be visually distinct from text, when examples are given in a paragraph (as contrasted with a display). That leaves intermediate forms in a tight spot if you give a derivation in the main text / footnotes, which then requires using bold or italics – which is forbidden by some publishers.

  • Yes, I never thought about it before, but I suppose one of the appeals for OT is that it obviates recourse to intermediary forms in this kind of derivation. Is that one of the reasons for its attractiveness In modern phonological literature? I wonder if these intermediary forms may have some sort of theoretical implications which I do not mean to be positing...
    – Teusz
    Nov 30, 2015 at 12:25
  • 1
    I don't think intermediate forms were at all relevant in the rise of OT, but that would be better for a separate question. Some people view derivations literally and hold that they are like the execution of an actual computer program with some number of nanoseconds separating each step of the computation. Mostly, we understand derivational stages as being about the logical flow of access to information.
    – user6726
    Nov 30, 2015 at 16:27

There's no standard and no way to say what is correct in linguistic theory. Linguists disagree about theory, and they always have. This talk about "standard theory" got started when Chomsky outlined some of his ideas and then said that henceforth he would refer to what he had described as the "Standard Theory". I have to think that he meant it as a sly joke. He proposes a name for his theory, for convenience of reference, as of course was his right, since it was his theory, but then, subsequently, anyone referring to this theory of his would have to call it "Standard", even if they just wanted to disagree with it. Get it? A pun on a proper noun. It's Standard (because that's its name), but that actually says nothing about whether it is standard.

No linguist was ever fooled by this cute little rhetorical trick, but it seems to have thrown a generation of linguistics students into deep confusion. No real linguist would ask the questions you've asked about what is "standard" and what is "correct".

Now, if you want to know whether what you've said about derivational stages in your example will make sense to phonologists, sure, it's perfectly clear. Appealing to intermediate stages of derivation was begun by Leonard Bloomfield, was controversial at the time (even for Bloomfield himself), became a routine part of Generative Phonology, then fell out of favor with some. But every phonologist knows how it works. That's not a problem. Do people agree with it? Some do, some don't, some don't care. Has it been designated as part of a Chomskian "Standard"? Who knows? Who cares? That's not a linguistic question.

As it happens, I don't personally believe in multiple steps of derivation, in either phonology or syntax. But I would never object to the sort of derivation you give in your example, because that sort of theoretical objection would presumably be beside the point of the analysis you're proposing. It would be irrelevant, and impolite.

If you're asking about style -- what is currently most fashionable among phonologists -- I don't know (or care).

I don't see anything wrong with your use of slashes and brackets. Generative phonologists routinely use slashes for underlying phonological forms and use brackets for phonetic transcriptions of any stage of a phonological derivation. What you have written seems quite clear.

  • On the Standard/standard pun, I've run across a classic essay about this and related matters by Ernest Nagel, and it happens to be available on line. For anyone who's interested, it's here: unz.org/Pub/NewmanJames-1957v03-01878
    – Greg Lee
    Nov 28, 2015 at 19:46
  • Thanks for the response, which is very helpful. You described that some phonologists find it problematic to posit intermediary forms in the underlying-to-surface derivation, and you yourself find it also somehow inappropriate (even if you'd accept it). Is this unease owing to some specific theoretical baggage that would be best avoided?
    – Teusz
    Nov 30, 2015 at 12:24
  • 1
    @Teusz, yes, it is due to specific theoretical baggage. Some years ago, there was a controversy between Fred Householder, who claimed that Generative Grammar was a theory "for the speaker", as opposed to a theory which describes what hearers do, and, on the other side, Chomsky and Halle, who argued (rather rudely) that Generative Grammar was actually neutral between speaker and hearer. Householder was right. In GG, the application of rules is conditioned by what is present in input forms, not output forms. This restriction make no sense, if the theory really is ... (cont.)
    – Greg Lee
    Nov 30, 2015 at 16:50
  • 1
    (cont.) ... to cover language perception as well as production. Once this restriction is lifted, intermediate phonological forms become unnecessary. (I've worked a lot on this, and I could tell you more, if you are interested enough to ask a question about how to avoid intermediate forms.)
    – Greg Lee
    Nov 30, 2015 at 16:53

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.